I am new to IPFS.
A project we are currently working on, has a network of 2000+ nodes running IPFS version 0.6.0
The network has a few clusters - one located in AWS and other clusters in 2-3 other private data centers; the gateway of each cluster, serves as a bootstrap node and there is yet another common bootstrap node connecting all nodes in the network.
The problem we have is that, the ipfs swarm peers results are inconsistent. The gateway node shows between 1.8k to 2k+ peers when we run swarm peers. One moment it shows 2300, the next moment 1500 peers.
The participating nodes within the Clusters show only 800 to 1k peers.
Nodes in AWS Cluster, (which has around 250 nodes) show only 7-12 peers. AWS nodes do not see each other, but they are seen by nodes in other clusters.
Trying to make sense of the above and identify what is missed in the network setup.
Any guidance to help to us ensure all peers are seen by each other at all points of time, would be very much appreciated
Check that connection manager configuration for Highwater and Lowwater are what you expect them to be.
Also, check that nodes can be dialed from other places. It may be that your AWS nodes cannot be reached from the outside and only have the connections that they initiate.
Thanks for the quick reply,
Setting up Highwater & Lowwater values in config indeed helped, to ensure the peers in gateway nodes maintain peer list well. Thank you again
However, the issue with nodes within the clusters (Excluding gateway nodes) and aws nodes still persist.
Actually, its the other way round, other cluster nodes can see AWS nodes; but AWS nodes cannot see each other. An ipfs ping or a swarm connect to each other within aws fail.
Well you probably need to open up ports in the security groups etc?
Thank you, that worked , great help